FraudGuard AI - Real-Time Fraud Detection Dashboard

A production-ready fraud detection system with real-time monitoring, 3 ML models, and a stunning React dashboard.

✨ Features

🎯 3 Fraud Detection Models

Basic Rules Engine (v1.0.0) - Threshold-based fraud detection
Machine Learning Model (v2.1.0) - XGBoost with 92% accuracy
Deep Learning Model (v3.0.0) - LSTM neural network with 96% accuracy

📊 Real-Time Dashboard

Live transaction monitoring with WebSocket
Interactive transaction analysis
Real-time statistics and risk scoring
Model performance comparison with charts

🚀 Production-Ready

FastAPI backend with WebSocket support
React + TypeScript frontend with Chart.js
Realistic transaction simulator
Model versioning and activation
Performance tracking and metrics

🖼️ Screenshots

Live Dashboard - Monitor transactions in real-time

Stats cards showing total, flagged, and blocked transactions
Live transaction feed with risk indicators
Detailed transaction analysis panel

Model Comparison - Compare model performance

Model performance cards with key metrics
Accuracy trends over time (line chart)
False positive rates comparison (bar chart)
Comprehensive comparison table

🛠️ Tech Stack

Backend

FastAPI - Modern async Python web framework
WebSockets - Real-time bidirectional communication
scikit-learn - ML model framework
XGBoost - Gradient boosting for ML model
TensorFlow/Keras - Deep learning framework
Faker - Realistic data generation

Frontend

React 18 - UI library
TypeScript - Type-safe JavaScript
Vite - Fast build tool
Chart.js - Interactive charts
React Router - Client-side routing

🚀 Quick Start

Prerequisites

Python 3.10+
Node.js 18+ and npm
Git

1. Clone Repository

git clone https://github.com/deymohit02/AI-Based-Fraud-Detection.git
cd AI-Based-Fraud-Detection

2. Backend Setup

# Navigate to backend directory
cd backend

# Create virtual environment (recommended)
python -m venv venv

# Activate virtual environment
# Windows:
venv\Scripts\activate
# Linux/Mac:
# source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

# Copy environment example
copy .env.example .env  # Windows
# cp .env.example .env  # Linux/Mac

# Start backend server
python main.py

Backend will be available at http://localhost:8000

3. Frontend Setup

Open a new terminal window:

# Navigate to frontend directory
cd frontend

# Install dependencies
npm install

# Start development server
npm run dev

Frontend will be available at http://localhost:5173

📖 Usage

Starting the System

Start Backend (Terminal 1):
```
cd backend
python main.py
```
Start Frontend (Terminal 2):
```
cd frontend
npm run dev
```
Open Browser: Navigate to http://localhost:5173

Using the Dashboard

Live Dashboard Tab

Click "Start Monitoring" to begin real-time transaction stream
Watch transactions appear in the feed
Click any transaction to see detailed analysis
View model predictions and risk assessment
Click "Stop Monitoring" to pause the stream

Model Comparison Tab

View performance metrics for all 3 models
Compare accuracy trends over time
Analyze false positive rates
Click "Activate" to switch active model
Review the comparison table and performance analysis

🎯 How It Works

Transaction Flow

[Transaction Simulator] 
    ↓
[WebSocket Connection]
    ↓
[Model Manager] → Runs all 3 models in parallel
    ↓
[Active Model] → Makes blocking decision
    ↓
[Dashboard] → Display results in real-time

Models

1. Basic Rules Engine (v1.0.0)

Algorithm: Rule-based threshold detection
Accuracy: 85%
Features:
- High amount detection (>$1000)
- Transaction velocity checks
- V-feature anomaly detection
Use Case: Fast, explainable decisions

2. Machine Learning Model (v2.1.0)

Algorithm: XGBoost (Gradient Boosting)
Accuracy: 92%
Features:
- 30 features (Time, V1-V28, Amount)
- Pre-trained on credit card data
- StandardScaler normalization
Use Case: Balanced accuracy and speed

3. Deep Learning Model (v3.0.0)

Algorithm: LSTM (Long Short-Term Memory)
Accuracy: 96%
Features:
- Sequential pattern detection
- 64-unit LSTM layer
- Dropout regularization
Use Case: Highest accuracy, complex patterns

Transaction Simulator

Generates realistic transactions with:

Merchant names (using Faker library)
Locations (cities and states)
Amounts (log-normal distribution)
V-features (PCA components, V1-V28)
Fraud patterns (12% fraud rate)
- High amount frauds
- Unusual locations
- Rapid transaction sequences
- Unknown merchants

📁 Project Structure

AI-Based-Fraud-Detection/
│
├── backend/                  # FastAPI Backend
│   ├── models/              # Fraud detection models
│   │   ├── rules_engine.py  # Basic rules model
│   │   ├── ml_model.py      # ML XGBoost wrapper
│   │   ├── dl_model.py      # DL LSTM model
│   │   └── model_manager.py # Model orchestrator  
│   ├── main.py              # FastAPI application
│   ├── transaction_simulator.py  # Transaction generator
│   ├── requirements.txt     # Python dependencies
│   └── .env.example         # Environment template
│
├── frontend/                # React Frontend
│   ├── src/
│   │   ├── components/      # Reusable components
│   │   ├── pages/          # Dashboard pages
│   │   ├── hooks/          # Custom React hooks
│   │   ├── types/          # TypeScript types
│   │   ├── App.tsx         # Main app component
│   │   └── index.css       # Design system
│   ├── package.json        # Node dependencies
│   └── vite.config.ts      # Vite configuration
│
├── models/                  # Pre-trained models
│   ├── xgboost.pkl         # Trained XGBoost model
│   └── scaler.pkl          # Feature scaler
│
└── README.md               # This file

🔧 Configuration

Backend Configuration (`.env`)

# Server
HOST=0.0.0.0
PORT=8000
DEBUG=True

# Model Paths
XGBOOST_MODEL_PATH=../models/xgboost.pkl
SCALER_PATH=../models/scaler.pkl

# Transaction Simulator
TRANSACTION_RATE=2.0     # Transactions per second
FRAUD_RATE=0.12          # 12% fraud rate

# Active Model
ACTIVE_MODEL=ml_model    # rules_engine | ml_model | dl_model

🎨 API Documentation

Once the backend is running, visit:

Interactive API Docs: http://localhost:8000/docs
Alternative Docs: http://localhost:8000/redoc

Key Endpoints

WebSocket

ws://localhost:8000/ws/transactions - Real-time transaction stream

REST API

GET /api/health - Health check
GET /api/stats - Current statistics
GET /api/transactions - Recent transactions
GET /api/models - All model metrics
GET /api/models/comparison - Model comparison data
POST /api/models/activate - Activate a model
POST /api/predict - Predict single transaction

🧪 Testing

Manual Testing

Backend Health:
```
curl http://localhost:8000/api/health
```
Start Frontend: Open browser to http://localhost:5173
Test Flow:
- Click "Start Monitoring"
- Verify transactions appear
- Click a transaction
- Verify details panel loads
- Go to Model Comparison tab
- Click "Activate" on different model
- Return to Live Dashboard
- Verify active model changed

🚀 Production Deployment

Backend (Render, Railway, or similar)

cd backend
pip install -r requirements.txt
uvicorn main:app --host 0.0.0.0 --port $PORT

Frontend (Vercel, Netlify, or similar)

cd frontend
npm run build
# Deploy 'dist' folder

Important: Update WebSocket URL in frontend code to match your backend URL.

📊 Performance Metrics

Model	Accuracy	Precision	Recall	F1 Score	FPR	FNR
Rules Engine	85.0%	88.0%	92.0%	90.0%	12.0%	8.0%
ML Model (XGBoost)	92.0%	94.0%	96.0%	95.0%	6.0%	4.0%
DL Model (LSTM)	96.0%	97.0%	98.0%	97.5%	3.0%	2.0%

🤝 Contributing

Contributions are welcome! Please:

Fork the repository
Create a feature branch
Make your changes
Submit a pull request

📄 License

MIT License - See LICENSE file for details

🙏 Acknowledgments

Dataset Inspiration: Kaggle Credit Card Fraud Detection
Tech Stack: FastAPI, React, Chart.js, TensorFlow, scikit-learn
Design: Modern dashboard UX patterns

📧 Contact

Maintainer: Mohit Dey
GitHub: @deymohit02
Repository: AI-Enhanced-Fraud-Detection

Built with ❤️ for production-ready fraud detection

Name		Name	Last commit message	Last commit date
Latest commit History 46 Commits
Project Reports		Project Reports
Synthetic Financial Datasets For Fraud Detection		Synthetic Financial Datasets For Fraud Detection
backend		backend
creditcard		creditcard
database		database
frontend		frontend
ieee-fraud-detection		ieee-fraud-detection
models		models
src		src
static		static
tests		tests
.gitignore		.gitignore
BalanceDataset.py		BalanceDataset.py
DeepLearningModels.py		DeepLearningModels.py
Dockerfile		Dockerfile
GIT_FIX_GUIDE.md		GIT_FIX_GUIDE.md
MachineLearningModels.py		MachineLearningModels.py
README.md		README.md
analyze_data.py		analyze_data.py
analyze_v_features.py		analyze_v_features.py
best_fraud_vector.json		best_fraud_vector.json
check_dataset.py		check_dataset.py
docker-compose.yml		docker-compose.yml
download_kaggle_dataset.py		download_kaggle_dataset.py
find_best_fraud.py		find_best_fraud.py
list_gemini_models.py		list_gemini_models.py
render.yaml		render.yaml
requirements.txt		requirements.txt
send_fraud_json.py		send_fraud_json.py
test_api_local.py		test_api_local.py
test_api_with_real_data.py		test_api_with_real_data.py
test_local.py		test_local.py
test_model.py		test_model.py
test_output.txt		test_output.txt
test_predictions.py		test_predictions.py
test_proper_vs_improper.py		test_proper_vs_improper.py
train_models.py		train_models.py
train_optimized.py		train_optimized.py
train_simple.py		train_simple.py
verify_fraud_logic.py		verify_fraud_logic.py

deymohit02/AI-Enhanced-Fraud-Detection

Folders and files

Latest commit

History

Repository files navigation